Forecasting Electricity Consumption of Electrical Machines of a Coal Industry Enterprise Using the Wavelet Transform

Electricity consumption forecasting is an integral part of the workflow in most industries that does not lose its relevance. Modern international energy-control systems do not function without forecast models of electricity consumption. At the same time, daily load schedules of the mining industry are characterized by deterministic chaotic fluctuations in the electricity consumption curve caused by the technology of the production process. The main criteria are the accuracy of the forecast values and the trainability of this model based on the actual data on electricity consumption. A high-quality forecast of electricity consumption for a given planning horizon can significantly increase the energy efficiency of enterprise operation in real time with the possibility of production management of an industrial enterprise by taking into account the possible conditions of the object. In the paper, a mathematical model of short-term forecasting based on the wavelet transform for each hour for the day ahead is proposed. The selection of the model structure taking into account the features of the described process help to find and work with noise and outlier values in the data. The forecast values obtained using eleven basis wavelet functions were compared, and the average percentage values of relative forecast errors for each hour were obtained. The predicted values of electricity consumption for a given planning horizon were verified on test data in order to assess the relevance of the forecast model.